Understanding and predicting pen fouling, tail biting, and diarrhoea in farmed pigs.
Domun, Y. (2023) Understanding and predicting pen fouling, tail biting, and diarrhoea in farmed pigs. Doctoral thesis, Harper Adams University.
|
Text
Yuvraj Domun Phd_Thesis_Corrected_Submitted - submitted.pdf Download (4MB) | Preview |
Abstract
This PhD research investigated precision animal farming, specifically emphasising commercially reared pigs and their welfare, addressing concerns like pen fouling, tail-biting, and diarrhoea. While animal welfare in pig farming is critical, there is a lack of comprehensive predictive models that integrate various factors affecting pig behaviours. The primary objective was to create advanced algorithms and predictive models that combine mechanistic modelling and machine learning to better understand and predict pig behavioural dynamics related to welfare issues. Various methods were employed, including transfer function models to link water consumption with temperature differences, analysing spatial positioning in relation to fouling events, and employing neural network architectures for time series data. Bayesian networks were utilised for simulating intervention scenarios. Several significant discoveries were made during the research. Anomalies in pigs' water consumption that were linked to temperature variations were effectively identified by the transfer function model, giving valuable insights into pen fouling and tail-biting incidents. It was also discovered that a crucial role in influencing fouling events in pigs is played by spatial positioning and temperature differences between different activity areas within pig pens. Superior predictive capabilities for events such as fouling, tail-biting, and diarrhoea were demonstrated by the innovative application of a neural network approach to predict these events. Furthermore, an early warning system that utilised hierarchical clustering and principal component analysis was introduced, which showed strong predictive potential. Finally, this research also demonstrated that Bayesian Network simulations can be used as a non-invasive method to test for potential strategies to mitigate welfare issues in farmed pigs while also providing practical insights for better farm management. This research offers vital tools and insights for advancing precision pig farming, fostering a more sustainable and ethical approach. The developed algorithms not only contribute to better pig welfare but also enhance monitoring, potentially leading to increased farm profitability. While the models are promising, further refinement and research into the various factors affecting pig behaviour are recommended.
| Item Type: | Thesis (Doctoral) |
|---|---|
| Divisions: | Engineering |
| Depositing User: | Mrs Rachael Giles |
| Date Deposited: | 14 Dec 2023 15:46 |
| Last Modified: | 14 Dec 2023 15:46 |
| URI: | https://hau.repository.guildhe.ac.uk/id/eprint/18034 |
Actions (login required)
 |
Edit Item |